Smoking Ban and Paediatric Hospital Admissions for Respiratory Tract Infections in England

Primary research question:

Has the rate of hospital admissions for acute respiratory tract infections (RTIs) among children aged 14 years and under decreased following the 1 July 2007 introduction of a ban on smoking in public places in England?

Study design:

Interrupted time series design

Study population:

The population at risk or is the number of children aged 14 years and under living in England at any time during the study period. The age restriction is applied to minimise the potential effect of self-smoking on the outcome.

Intervention:

The intervention under study is the ban on smoking in enclosed public places implemented in England 1 July 2007.

Study period:

The study period is 1 January 2001 to 31 December 2012. This is the maximum time period surrounding the ban's introduction for which both denominator (population at risk) and numerator (number of hospitalisations) data are available through the data sources at the geographical area level required.

Outcome:

The primary outcome is the rate of unplanned hospital admissions for acute RTIs. Secondary outcomes include the rate of unplanned hospital admissions for acute upper RTIs (URTIs), and the rate of unplanned hospital admissions for acute lower RTIs (LRTIs). Admissions containing both a diagnosis of an URTI and a LRTI will be counted as a LRTI only. The secondary outcomes are thus mutually exclusive.

The following International Classification of Diseases (ICD)-10 codes will be used to identify acute RTIs:

URTIs: A37, H66-H67, J02.0, J00-J06, J09-J11 (excluding J10.0, J11.0) LRTIs: J10.0, J11.0, J12-J18, J20-J22, J40-J42 Only unplanned hospitalisations where either a primary or first secondary diagnosis of an acute RTI is recorded, will be included. Admissions with a primary diagnosis of asthma are excluded to prevent overlap with a previous study assessing the impact of the English smoking ban on paediatric asthma hospitalisations. Transfers between hospitals following initial admission will not be included. As a unique patient identifier is not available in the source database, it is not possible to distinguish between first and subsequent admissions for individual children.

Data sources:

Data on the number of unplanned hospital admissions for acute RTIs will be derived from the Hospital Episode Statistics (HES) database governed by the Health and Social Care Information Centre (HSCIC). HES contains individual-level data on all hospital admissions at National Health Services (NHS) hospitals across England. Mid-year population estimates to define the population at risk will be obtained through the Office for National Statistics (ONS).

Data extraction and handling:

The number of unplanned hospital admissions for acute RTIs among children aged 0-14 years as extracted from HES will be aggregated by HSCIC into strata based on all possible combinations of the following covariates: age group (0-4 years; 5-9 years; 10-14 years), sex (male; female), Middle Super Output Area (MSOA), admission month, and admission year. MSOA is a geographical indicator for areas with 5,000 to 15,000 inhabitants. A count variable indicating the corresponding number of hospitalisations will thus be assigned to each stratum.

Mid-year population estimates according to strata based on all possible combinations of the same covariates will be extracted. Monthly population estimates will then be calculated via linear extrapolation of mid-year estimates.

The following additional covariates will be obtained by linking the specific indicators at MSOA-level: region, urbanisation level, and socioeconomic status (SES). MSOA will then be dropped as a covariate and counts will be aggregated into strata based on combinations of the following covariates: age group, sex, region, urbanisation level, SES, admission month, and admission year. RTI hospitalisation rates will be calculated for each stratum.

Sample size:

Sample size calculation for time-oriented analyses is complicated given the complexity of the models, and in a way redundant given that nationwide data are being used for the current study. No prior studies have specifically evaluated changes in the number of hospital admissions for respiratory infections among children following the introduction of smoke-free legislation.

In a previous epidemiological evaluation of the Scottish smoking ban a highly significant (p<0.001) annual drop in asthma hospitalisations of 18% (95% confidence interval (CI) 15-22) was found among children <15 years of age. Meta-analyses of observational studies indicate that second-hand smoke exposure is associated with higher risk of respiratory tract infections in infancy (OR 1.54 (95% CI 1.40-1.69) for lower respiratory tract infections in infancy and 1.62 (95% CI 1.33-1.97) for middle ear disease in childhood) when compared to incident and current asthma in children (OR 1.21 (95% CI 1.08-1.36), and 1.30 (95% CI 1.22-1.39), respectively. This indicates that smoke-free legislation is likely to have a larger effect on acute respiratory infections than asthma. Furthermore, paediatric hospital admissions for acute respiratory infections are more common than for asthma, and the study period for the current study is longer than compared to the previous asthma study. Therefore, the current study is expected to have ample power to detect a significant and clinically relevant drop in hospitalisations for respiratory infections, if present.

Statistical analysis:

Relevant population determinants will be described for the pre and postban periods. Aggregated hospitalisation rates will be modelled over the study period using an interrupted time series approach. Changes following the introduction of smoke-free legislation will be studied, taking into account seasonal patterns in RTI rates and the potential effect of relevant covariates (age group, sex, region, urbanisation level, SES). Final model selection will be based on the corrected Aikaike's Information Criterion (cAIC).

On 4 September 2006 the 7-valent pneumococcal vaccine (PCV) was introduced into the childhood immunisations schedule at 2, 4, and 13 months of age, with a catch-up programme for children born from 5 September 2004. The relative contribution of true pneumococcal infections to the total burden of admissions for respiratory infections, the majority of which is likely to be of viral aetiology, is expected to be small. However, considering the close temporal proximity of PCV introduction to that of the smoking ban a sensitivity analysis will be performed to study its potential effect on the impact estimation of the smoking ban.

All analyses will be performed for both the primary and the secondary outcomes using Stata 12.0.